Hello everyone,
Could anyone shed some light on this matter?..... Minitab gives an example
for the ungrouped case in binary logistic regression. Pearson's statistic
and the Deviance statistic are quoted but with degrees of freedom equal to
47. In this case the number of observations is N=92 and the model has p=3
parameters. Usually df=N-p. Does anyone know why it is 47 in this case?
Perhaps it is due to the data being of the ungrouped type and Minitab
perhaps have used 'modifications'? If so, does anyone have a reference for
such modifications?
Many thanks for our help,
Kim
|